This is the second Reg article about attempts to improve the simulation of complex fluid dynamical systems in a week - the first was about using machine learning to determine where courser but faster simulations could be used in place of finer and slower calculations.
Good.